Reality Capture API on Forge – Build highly scalable reality capture applications

KARTHIK NATHAN: All right welcome, everyone. This is the reality capture on Forge talk. My name is Karthik, and this presentation was partly created by Murali, as well. And we have Dominique, who's sitting right in front. So if you have any questions, you can ask me, Dominique, anyone. So let's get started.

So yeah, in this presentation we'd like to take you through how Autodesk views reality capture and the future of making things. So we'll first talk about reality capture products, and then finally, we'll end up with a live demo of Forge API. So we are going live on Forge today. So you guys can start using these APIs as well. And we'll walk through a live demo.

Oops, wrong way. Let me play a real quick video, so you guys.

[VIDEO PLAYBACK]

- For any design or renovation project, you need to capture existing conditions. But traditional methods of surveying are painful, time consuming, and tedious. Multiple trips to the job site and manual capture methods can lead to schedules overrunning, and base information that's often incomplete or inaccurate. And inaccurate data costs time and money.

Fortunately, reality capture is changing the way we survey. It's a process that captures, measures, and maps millions of surface points to build a textured, high resolution, precise 3D model of your site. You'll be able to scan quickly and accurately, making it easier to survey, plan, renovate, and construct new buildings or infrastructure. You can scan sites faster and easier than ever before, using reality capture.

With ReCap Pro, we've simplified the steps so anyone can use it, whether you're a beginner or a specialist. Kickstart your project with reality capture in three steps. First, gather detailed data using lidar or photo photogrammetry technology. A camera mounted to a drone or a laser scanner can scan the site in hours, not days. Some laser scanners are now small and light enough to carry in a backpack, and they continue to reduce in cost. Next, ReCap Pro automatically registers, cleans up, and analyzes the data. The end result is a point cloud, or mesh, that is ready to use with CAD and 3D modeling software. Finally, once you've imported the data, you're ready to start designing, working with information that's up to date and accurately represents real world site conditions.

Reality capture is a simple time saving method for capturing and converting site information into accurate 3D models. Design and build with confidence.

[END PLAYBACK]

KARTHIK NATHAN: OK, so before we start, I just want to get a show of hands just to see who the audience is. Who's new to photogrammetry and reality capture? OK, quite a few of you. And whose technical, who knows like REST APIs? OK, quite a few as well. So that's good. So what I'll do is I'll spend a little time on photogrammetry, and ReCap, and ReCap Photo, and then we'll move into the demo part of it.

So what is ReCap? So ReCap is Autodesk's solution for reality capture. It's a hardware agnostic platform. So it doesn't matter what hardware you have, what scanner, or what camera you have. It'll work with all kinds of data. You can bring in photos. You can bring in laser scans. I can bring in lidar data, aerial photographs, and we'll be able to process that. And what does it give you? It gives you deliverables like a mesh, like you saw in the video, textured mesh, a point cloud, or ortho images that will enable you to take it to enable scan to BIM workflows.

So what are these used for? So who's using ReCap right now. So we've talked a lot of customers, a lot of them already have some kind of capture process in their workflows. So these are some of the industries that are using it, factories, plants, construction sites. All of them have one thing in common. They're all looking to measure change and monitor change. So what's changed in the project timeline. So we're looking to improve these workflows, simplify them, and make it a little more cost effective.

So this is one of the use cases, where you're actually preparing data to take it as BIM capable workflows. So ReCap Pro is our software that aggregates this reality capture data. And then from there, we process this data to create these ortho TIFFs meshes, point clouds, that you can then take into our hero products like AutoCAD, Revit, Civil 3D, and further.

Then there's the other side of photogrammetry and reality capture, where you just want to capture that as built data. This is mainly used for a multimedia and education and digital heritage preservation. And these are usually photogrammetry workflows, so with photos.

ReCap Photo is-- I don't know how many of you have heard of ReMake? OK, quite a few. So ReCap Photo is ReMake rebranded. And we made enhancements to it. I'll talk about those in a bit. ReCap Photo is primarily focused on aerial workflows, aerial photogrammetry workflows. But I'll talk about some of the enhancments as well. Here's a quick, nice video actually. I'll show you this real quick.

[VIDEO PLAYBACK]

[MUSIC PLAYING]

[END PLAYBACK]

So ReCap Photo is actually comes bundled with ReCap Pro. So if you buy a subscription to ReCap, you'll get a ReCap Photo as well. It's a freemium product. So a lot of the capabilities are free. The processing of the images, we charge based on cloud credits. I'll get to that later. It's a cloud connected app, so all the processing is done in the cloud. You don't need a heavy machine to process this data. It's also a mesh editing and mesh cleanup app. We can look through that as well later.

So photogrammetry for people who are new, I'll run through this very quickly. Photogrammetry is the technique of taking multiple images from different viewpoints, combining them, stitching them to create accurate 3D model. So the three big parts of it-- there are actually much more, but these are the three big parts. So the first part is actually structure from motion, where you actually determine the camera positions. And then from that, we can then-- it's called multi-view stereo, where we can then, using each pixel, we can get the 3D coordinate of every single point. And then from there, we can then get our deliverables like ortho TIFFs and our meshes.

Photogrammetry, so these are why we are talking about equipment here, because the quality of your input determines the quality of your output. So it's always important to be aware of the limitations and the hardware requirements for this that'll result in good output. So it's always good to use a high resolution camera and lens that we support. Rectilinear lens, fish eye is not supported. And we rely on the EXIF data for georeferencing. So the GPS greatly enables geolocating your data. This is primarily for aerial photogrammetry.

These are some of the things to keep note of. There's quite a bit here, but if I can sum it up, you need to have good overlap between your photographs. We recommend 60% to 80%, laterally, as well as forward. If you're capturing an object, so if you're not doing aerial photogrammetry, 60% of it must be in the frame in every single photo. It must be well lit, especially. So if it's outdoors, do not do it on a sunny day. Do it on an overcast day. So diffuse lighting is the best.

Photogrammetry, here's the limitation. It doesn't work well with repetitive patterns, reflective surfaces, shiny surfaces. So obviously surfaces with a lot of texture, a lot of color, a lot of detail will work great. But there are ways to get around this, I'll talk about this as well. The entire image must be in focus. So blur is not good. And yeah, you can use a turntable. If you plan to use a turntable, then the background must be all white or green, one color.

OK, so aerial photogrammetry. How does this tie into ReCap Photo? There are a lot of third party drone apps. So a lot of the drone manufacturers have their own apps, but you get a lot of them on the app store as well, as Android Google store. A lot of them are free, as well. So the drone apps automate a lot of this stuff for you, so you don't have to worry about it. You just mark up the area that you want to map, and it'll calculate where it needs to take the photos from. I'll also make some recommendations as to how much overlap you just have for different types of sites. So for example for a very flat agricultural land, it's recommend that you fly higher, at a high altitude. And if you're capturing a lot of buildings, then in addition to taking photos straight down, you'll also want to take a lot of obliques. So these drone apps will help you do all of that. Once you gather your photos, then that's where it starts feeding into Autodesk ReCap. So this is kind of what's going on.

So as I mentioned, ReCap Photo is a-- it's a desktop application, but it's cloud connected. So once you bring in your photos, the first thing that's done is it goes up into our cloud. And we use AWS back-end to process your data. Forge clients also will directly talk to our service, instead of going through ReCap Photo. As a result of that cloud service is what you get-- you have 3D surfaces, point clouds, and orthoviews, which you can then take into our hero products. We are continuously working to improve the quality of these, as well as improve the ease in which you can take it to Civil 3D, Revit, et cetera.

So we ReCap Photo is also-- it's a large mesh we use. So we generate a lot of data, and we are able to process this as well and edit them. So I'll show you a demo of that later.

So here, it's good to take photos looking straight down, but as I mentioned for buildings it's good to also take obliques. You want a 60% to 80% overlap, but as I mentioned, the drone apps will make a lot of the recommendations for you.

We also support geolocation. So if you have ground control points, if you have data that's collected from your high precision GPS devices, you can feed them in. I'll show that in the demo as well. We have a ground control point schema that if you follow it, your ground control points will be used. It's good to place them, spread them apart in your scene. So don't put them all into one area. Spread them apart. And as a result, your accuracy will greatly improve throughout the scene. If you have rtk and ppk, GPS data which is very high accuracy GPS data, we are able to process that as well. I'll show you how we can process that.

So that's a screenshot of ReCap Photo. We provide a UI interface where you can enter your GPS data. And in the upcoming update, you'll also be able to visualize it using a maps view in the app itself.

So these are some of the results that we got. So by processing 561 photos, we were able to capture 40 acres. And you can specify for your ortho TIFF you can specify your ground sampling distance. So you can go as low as one centimeter to one meter for your ground sampling distance. That means one pixel in your ortho TIFF will be either one centimeter or one meter. You can control that. You get a pretty good point cloud and a mesh, as well. So we'll look at live examples later.

And this is some of our objects for close range photogrammetry. So this is the statues in Singapore. I'm sure if you used the ReMake, you've seen this. You can either do hand-held or on a tripod. Sorry not a tripod, on turntable. So the second example is that using a turntable. And all these available on our website. So in our documentation, you can find recommendations on how to take photographs, and how not to. So for example here's a classic one interior. If you're capturing interiors, it's best to actually walk around the perimeter of the room, and take photographs of the opposite end, rather than standing in the center and turning around yourself. So these all available in our documentation online. So you can find these.

And this is, again, an object that we kept, the Garuda statue. So we processed 850 photos. We've got extremely high resolution mesh, a pretty big size, and we're able to handle them.

OK, any questions before I move into the Forge demo and APIs? OK, so our service is live. If you go to this website, you'll be able to find how to get started and documentation. But we'll walk through it real quick, the basic steps of creating an app and the four basic steps of creating a project, submitting images, and actually getting results. So I'm going to move to my computer, and - switch over to Forge.

OK, so if you go to a developer.Autodesk.com, so this is for the technical people. The first thing you will-- how many of you have created Forge app? Excellent, so you don't need to do this, but for those who haven't, you will see this very simple. You can create a Forge app. I've already created one. You can give it a name. You can give it a description. The callback URL, this is mainly for three legged authentication. So if you want your customers to use A360, then you would need to give this and your website URL if you have one. So I've already created one, so I'm not going to do that here. But what this will give you is, it'll give you a client ID and a client secret. So this is what you use for all your API calls. So once you've created the app, now you can start using Autodesk Forge APIs. And we'll look into specifically how to use reality capture APIs.

So I'm going to switch over to this tool called Postman to show you how we get started. Actually, before I do that. A few more things I can talk about here. So the few basic steps, I'll show you this. So first thing is capture the images. Second thing is you need to authenticate yourself. So that's the app client ID and secret I told you about. So we'll walk through those APIs as well. Once you're authenticated, the next thing is to create a project. So there's one API call to create a project. Once you create a project, you need to associate some images to that project. So there's another API just to add files to it, images and GCP data. Once you've done that, you can start processing. So again, it's very simple. One API to create a project, one API to add files, one API to start processing. You can get a progress API, and one final API to get your results. So four basic steps. We have many more APIs, but I'm not going to talk about it here. It's all on our documentation, like how do I cancel a project. How do I delete a project. How do I query for projects.

Authentication they are two main workflows. So one is the two legged authentication. So for example, what I'm going to show you is just a simple use case of two legged authentication. So your app can directly talk to our services. If you want to use your user's A360 account and their accounts, then you would need to get their authorization. So that's a three legged workflow. I'll show you those APIs as well.

So these are the basic APIs, so the authenticate API, you will provide your client ID and secret and you'll get an accen token that you can then use. For three legged, you provide your client ID, we'll redirect them to A360, that will return an authorization code, and return you call the callback that you had before. And then you can request to get an access token. That'll give you an access token and a refresh token. So the refresh token is-- I'm sure you know. You don't want to keep asking the user to sign in every time when the token expires. So the refresh token can help you refresh the token, so you can continue using. So photogrammetry is an compute intensive process. It does take time. So you would need to use these refresh tokens.

OK, to create a photo scene you would call the POST photo scene API. There's a link to the documentation. This will be shared later. There's some parameters you can provide. Obviously, the scene name, the project name. The scene type, if it's an aerial object capture. The format that you want, RCM is the mesh, our propriety mesh format. OBJ, you can of course run OBJ be as well. RCM's a point cloud, ortho, and report's actually automatically generated, but you can specify them as well. I'll show you an example of the report. Callbacks, if you want to be notified when the processing is done you can actually provide a HTTP callback or email callback, and you'll get an email when the processing is done.

For aerial photogrammetry, we have a few more additional parameters. You can specify if you're using a regular GPS that's built into your camera or your drone, or ppk, rtk data. So if you have high accurate data, you can provide that as well. We take that into account. So we give more weightage to your GPS information. You can also, as I mentioned, you can provide ortho GSD, your ground sampling distance. So anywhere from 1 centimeter to 100 centimeters. And we support geolocation and coordinate systems now. So whatever Civil 3D, and all our hero products support, we support as well. We have a link to supported coordinate systems you can provide that. You can query that and provide that to our API.

After creating a project, now I want to associate some files with this project. Again, a very simple POST file call. I need to just provide the photo scene ID. So in the previous, when I create a photo scene I get an ID for that project. I just provide that ID, and I provide the type of image and the file. So the type, I can actually provide images. Chances are most projects one won't have GCP data. So it'll just be images. . But if you have GCP data, then you have to provide the type as survey, so we know that it's not an image. It's actually GCP data. And again, we have a schema. I'll show you that as well. So if you follow that schema and provide that XML file to us we will be able to handle your GCP data.

And that's it. You're done. Now you're ready to start processing. Again, very simple one, your post photo scene with your photo scene ID. Your project has now started processing. And if you want to query for updates, again, you just query with your GET progress end point. And, again, all you need to provide are your photo scene ID, and you will get a response. The body will contain a progress message, whether it's in progress, it's completed, it failed, it succeeded, and also progress percentage. And when it's finally done, you're ready to get your data. So again, it's a GET photo scene with a photo scene ID and end point. You provide the project idea again, your photos scene ID. And since we generate multiple deliverables, you can get all of them separately or together. So you can provide the format that you want. Say I only want the ortho for now, I can request just for the ortho.

What it returns is a HTTPS link that you can download your data from. It's auto expire is right now, it's seven days. And once you get the link, you can just download it. And in seven days time it will automatically be cleaned up. Or if you explicitly call the delete end point, which is I don't have in this presentation, but it's on the documentation. If you called delete, it will delete it straight away.

So before I get on the demo, this is how much it's going to cost you, right, API pricing. So we charge Cloud Credits, again. 1 giga pixel is 3 and 1/2 Cloud Credits. So 1 megapixel processed is 0.0035 Cloud Credits. So these are like roughly how much it'll cost you. So yeah, 1 giga pixel is quite a lot.

OK, let me get back into the demo now. So I've prepared some scripts, but I'll show you real quickly. So what I've done is I've gone ahead and I've prepared some end points. So let me show you this. OK, so as I mentioned, the very first thing you need to do is authenticate yourselves. So this is my authentication endpoint. The base URL is nothing, but I've stored it, so it's in a variable here. But it's nothing but developer.API.autodesk.com And then authentication is our end point. And as you can see, in my body I've provided my client ID and my client secret.

Scope is important. So scope is what kind of privileges you want. So I've said you need to have data read and data write permissions. So if you request for that, it'll return either successful or failure. In our case, you need both. So ask for these permissions, and then once you call it-- let me send this. So can you guys see fine?

So the response I get, I get an access token, and I get how long it's valid for. So it expires in 3,600 seconds. And with this access token, now I can make every-- every call I make after this, I need this access token to use this. So once I've authenticated myself, using my Forge credentials, I can now start creating my reality capture project. So again, the base URL is my developer Autodesk.com, the Forge API. And point and then this is the photo scene API it's talking about. Again, all I'm providing is the bearer and access token that I got from the previous call. I'm providing that.

In my body, we talked about the different parameters, right? So here I'm giving it a scene name. I'm Karthik, So I've called the project Karthik. And this is a statute. I only want to the RCM and OBJ for this. So I'm requesting for-- this is a comma separated list, so I'm requesting for an RCM and OBJ. So all of these are optional, actually. So the default format is just the RCM, if I don't provide anything. And scene type, by default, it's aerial. But in this case, I know I'm processing objects and providing that. And for callback I'm just going to give my email. So after it's done, it will just send me an email saying it's done. So if you have an app, you can provide an HTTP callback that your app can handle and take care of it from that.

So if I call this using my token that I gave, it returns me everything was fine. The usage is just how long it took. And it gives me a project ID. I can use this project ID to continue further.

And now I'm ready to start adding images. So again, I will call my final end point. In my header, again I'll provide the access token. In my body, here I have to provide the project ID, the photo scene ID did that I got from creating a project. So I provided that. Again, I'm using variables here so I don't have a copy paste. In this case, I'm saying him providing images. So I provide the type is image. And here I've already uploaded my images to an Amazon bucket. I have my own bucket, so I've put my images there. And I'm just providing links to those images. And I'm using an array file with indices.

You can technically put all your 500 images in one API call. We recommend you don't do it for now because your request will likely time out. So we recommend grouping it in 10 or less. So, for example here, I've made one API call. As soon as I'm done processing these 10, I can then process another 10, and these will have new links. I'm using the same zero, too. So I can call this file end point multiple times, as many times as I want. And I've grouped it in 10, or you could do individually as well, if you want. If you don't have your files in S3 or some public link, you can directly upload from your hard disk. That's also supported. So you don't need to put it somewhere else.

So let me run through this. Let me send this. So what it's going to do, it's going to add these files from these locations to my project. And you can see from the response, let me expand this little. So you can see it's given me details for every single image and no error. And it's uploaded successfully.

I can do the same with my second set. So nothing different here. Same thing, body's same, type is the same. The only thing different are images themselves. And again, I can send that. So what I've essentially done is I've created a project, and I've added 21 images to this project. So this is a pretty tiny project. It should finish quite fast. I'm not going to wait for it to finish, but I'll just show you. So again, it returns the JSON with everything successful, but if there was an error, it'll tell you error. Error codes are, again, in our documentation online. I'll show you that again later.

And that's it. I've added files, and I'm ready to start processing. It's super simple, again. My POST photo scene end point, I provide the project ID that I want to start. Again, my access token, and there's no body. And with this, if I send this, there you go. So it started, and it's no error. This is my project ID. So now this is where I need do a little copy and pasting.

So if I take this project ID, if I want to check on progress, so here's my GET progress end point, progress. Actually, I don't need to copy since it is the last one. So if I send this, there you go. So it says my project's created, progress is at 0%. I just started it, but over time it'll give you the accurate progress. And once it's done, this should say 100%. And you'll remember we provided a callback, so this should give you an email or HTTP callback. I'll show you that what that looks like.

So this is a very simple email that you get that it's ready, and now you can go get your results. Once you know it's complete, if you want to get your results, all you have to do is call a photo scene with the project ID. And here you notice in my query parameters, I've specified RCM. If I send this, I should get response saying hey, it's done. So this is obviously not the project I just started because that's still processing. This is a project that I started yesterday.

So it's done, progress is 100%, and it gives me a link that I can just then download directly from. And the same thing for OBJ if I'd asked for an OBJ. RCS, if I'd ask for an RCS report. So it gives me, yep it's all done, and here's my link.

Yeah, and it's as simple as that. You can even, as I mentioned, I didn't show it in my presentation, but you can cancel a project. If I call this hit Send Now, it'll actually-- and my photo scene ID is saved. So what I just started, it'll cancel that project. So if I don't want to continue processing. So one thing to know about charging, if your project fails or it succeeds-- sorry, if it succeeds, you'll get charged at the end of processing. If it fails for some reason, then you won't get charged. If you cancel, however, so if you initiated the cancel, we still charge you for the processing. Again, details are all on the website.

So yeah, this is the developer web site, the Forge documentation. So it walks through the basic steps, and also gets into the API details. So whatever I showed you, it has a lot more detail on the API endpoints. It also has-- I mentioned the GCP schema, right, the ground control points. So if this loads. So this may look complex, but it's not really. You just need to provide the coordinate system that your ground control points are in, and you provide the GCP coordinates, longitude, latitude, altitude, and as well as the images that they are on, and the pixel locations. So ReCap Photo, our app, our client provides a UI to do this. But if you have your own app, you can provide your own UI for the user to add GCP ground control points. Once you generate this and give it to us, we'll process it.

So this is an example of a report that we generate. So we give you details like, OK it took 55 minutes to process this. There were 70 images. We were able to calibrate all 70 of them and add GPS information. It gives you the ortho photo dimensions, as well as the texture size. We also provide a digital elevation model. So I'll show you that in a second. And we give you some statistics as to how well it stitched and how much overlap there is in the photos.

So in this case, if I use a ground sampling distance of one centimeter, your ortho photo is going to be humongous, because each pixel is one centimeter. So if I'm capturing one acre, you'll get that many pixels. So depending on the accuracy or the resolution that you want, you can specify your GSD. So if you're looking for a very high resolution, one centimeter is as good as you can get. Typically, 5 to 10 is also OK. So this one, I've used 10 centimeters per pixel.

And if you have ground control point data, it actually gives you some information on that. So you can even use check points to validate. In this particular example, I did not use checkpoints. I just used ground control point. So you can see the error is actually less than two pixels. So this is less than 20 centimeters, less in 0.2 of a meter. So the error is actually less than two pixels. And your check points, if I had checkpoints, you'd be able to see it there as well.

Here we go. So this is the point cloud that we generate, and this is Civil 3D. Once you bring in your point slide into Civil 3D, it's already located. So if you turn on it Bing Maps overlay, you'll see it properly geolocated. You can even bring in your ortho TIFF file and your point cloud, your TIFF, ortho TIFF, and Bing Maps all will align perfectly. You can do measurements here and stuff. So let me just show you orbit.

So this is a skate park in Singapore. I'm from Singapore, so I captured this. This was about 200 photos, about 500 meters by 500 meters, or 400 meters, something like that. It took about two hours to process. So two and a half hours to process.

This is ReCap Photo. So this is the mesh that we generate. So here's a little trick. So I can actually query for points. So if I want to know how accurate this is-- sorry, let me zoom out a little. I can actually click on a point. It'll tell me the geolocated coordinate of that. I can even do measurements. So if I go here, I want to know what's the distance between here and here. 17.865 meters.

So this is just a very simple example, but obviously have markers in your capture. Then you can actually do accurate measurements, but I don't have any markets in this.

And that is just to complete this. This is the ortho TIFF. So it's quite big. As you can see it's, 288 megabytes. So I used one centimeter per pixel. So if you use 10 centimeters, you'll be a lot smaller. So now you can do all kinds of measurements on this.

OK, I believe that is-- yep, that is the end of my presentation. If you guys have any questions technical or not, feel free to ask. Yes, sorry, I saw you first.

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: So the question was, do we handle video? Right now we don't, but if you take snaps of the video, and they have the EXIF that we need like GPS. If you're not interested in GPS, if it's not aerial, then it's not needed. But we don't recommend it because you do have motion blur and stuff, though, which is get you good results. Yeah, sorry.

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: Yeah so, a lot of things happened. So first is we have a new stitching engine. So our results are much better, accuracy is much better. That's internal. And we have tuned it for the aerial photogrammetry use cases. So buildings with sharp edges, those are retained now. So we have a much better engine. Now that's under the hood. And we have exposed this as a REST API. So we're going to make, as I said, we're going to make continuous changes on the back end, as well as on the API layer. So the APIs will become easier to use over time, actually very soon. And the engine will also continuously improve.

Yes, right now we support 1000 photos for aerial. And same thing, ReCap Photos is client as well. We support 1000 photos in that. Four objects, right now we are supporting 300 photos, but again we are working to improve that. There was a question in the back I think sorry. You had a question.

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: Sorry?

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: Model scene as in?

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: Oh, not yet. Not yet. But we are working to improve those workflows so we can. Yes?

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: So right now, it is laser scan data or photo data. You can go from a laser scan to a mesh. And the other way around sorry.

MURALI PAPPOPPULA: So I can speak for this one ReCap Pro [INAUDIBLE] point The goal here is to have [INAUDIBLE] scanning that you can [INAUDIBLE] read it or work out for.

KARTHIK NATHAN: So some of the other improvements we're making, so right now you when you're getting a results you get a link that you download your data from. What we're trying to do is make it easier for that data to flow into the scan to BIM workflow and the hero product. So we will closely tie this into the users Forge-- sorry A360 account. So it'll almost be a user starts and project, creates a project. Once the data is ready, if he opens up BIM 360 doc, for example, it'll automatically show up there. He doesn't have to download it and take it to there himself or herself. Any other questions yes?

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: Your neighbors property.

AUDIENCE:

KARTHIK NATHAN: So that one you will have to rely, have you used any drone apps to do your captures?

AUDIENCE: Some [INAUDIBLE].

KARTHIK NATHAN: I think that is probably the only way, right

MURALI PAPPOPPULA: Yeah this is difficult because Either you fly multiple images. It might be on the sides you capture something that belongs to the neighbor. Or you use obliques. When you move around the house, but when you have oblique images [INAUDIBLE] rest of the [INAUDIBLE]. So yeah there is no easy answer for this one. [INAUDIBLE] What you can do then is probably [INAUDIBLE] Actually needs the source photos anymore.

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: Any more questions?

AUDIENCE: [INAUDIBLE]

KARTHIK NATHAN: So it we use our Forge APIs right now, no. You'll get your OBJ, but very soon we're going to enable that. So the moment your outputs are generated, you don't have to do anything. They'll automatically show up in the large model of viewer of Autodesk and also BIM. Yeah, main docs.

MURALI PAPPOPPULA: Just a small comment. The only [INAUDIBLE] the OBJ will not be [INAUDIBLE] Otherwise [INAUDIBLE]

AUDIENCE: Can we pass in the geolocation?

KARTHIK NATHAN: Yeah so you'll have to do that yourself. So the OBJ file format doesn't support it and neither does LMV. So but RCM, our mesh format does, and all point cloud, ortho TIFFs, all geolocated. So if you open them in Civil 3D or whatever, it'll be geolocated. The OBJ, if you want to geolocated that, then you'll have to manually translate that to the coordinate that you want. Any other questions?

OK, so I'll make these available so you can take a look, and you'll have all the links, and you can give it a try. Yeah, if you have any questions or anything else, you can shoot us an email.

MURALI PAPPOPPULA: [INAUDIBLE] ReCap Photo which is using exactly the same API, we would be running ReCap Photo visuals as [INAUDIBLE] of the exhibit area. So there would be a [INAUDIBLE] and there will be a few kiosks. So I would be will. Karthik will be there. We can [INAUDIBLE]

KARTHIK NATHAN: All right thank you all for coming.

[APPLAUSE]